: For most users, the simplest way is to install the latest version from threadpoolctl on PyPI : pip install threadpoolctl Use code with caution.
git clone https://github.com cd threadpoolctl pip install -e . Use code with caution. Key Features and Use Cases
: If you use Anaconda or Miniconda, install it from the conda-forge channel via Anaconda Cloud : conda install -c conda-forge threadpoolctl Use code with caution. threadpoolctl download
: You can inspect the thread-pool status of imported packages directly from the terminal: python -m threadpoolctl -i numpy scipy.linalg Use code with caution. Practical Example: Limiting Threads in Code
You can use threadpoolctl as a context manager to restrict thread usage within a specific block of code: : For most users, the simplest way is
: The ThreadpoolController class allows users to filter and select specific libraries to limit while leaving others untouched.
: Developers can clone the official GitHub repository and install it in developer mode: Key Features and Use Cases : If you
: Dynamically sets the maximum number of threads these libraries can use. This is vital when running parallel loops (e.g., using joblib ) where each worker might otherwise try to use all available CPU cores.
from threadpoolctl import threadpool_limits import numpy as np # Limit BLAS libraries to use only 2 threads for this specific operation with threadpool_limits(limits=2, user_api='blas'): # Heavy linear algebra operations here res = np.dot(np.random.randn(1000, 1000), np.random.randn(1000, 1000)) Use code with caution. Compatibility and Requirements threadpoolctl - PyPI
You can download threadpoolctl using several package managers depending on your environment: